iSixSigma

Gage RR for Discrete ( Pass / Fail ) ?

Six Sigma – iSixSigma Forums General Forums Tools & Templates Gage RR for Discrete ( Pass / Fail ) ?

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #53682

    Hicks
    Participant

    We have 5 operatiing systems ( K-2, Adage, Oracle, Red Prairie, Whistle ) that contains product specs ( Ti-Hi, case weight, cases/pallet, case id #, etc…. ).

    I am trying to quantify the differences between the 5 systems.

    Example:

    Item # 15299 ( spec should be 10 cases per layer / 10 high = 100 cases per pallet ).

    I want to check the 5 systems to ensure all 5 have the same information.
    I was trying to relate it to a Gage RR ( or similar ) to show differences – between programs and within programs ( considering I have numerous 10/10 item #’s ).

    Any suggestions on the tool I should use and a template?

    Thanks.

    0
    #191092

    MBBinWI
    Participant

    Alan: Not really a GR&R. Are you extracting the data electronically, or do you need to have people go in and audit the data? If the latter, you might want to conduct an MSA on their ability to reach the same conclusions (attribute study).

    If you can pull the data electronically and put it into a count type format, perhaps analyzing via Chi-Square would help.

    0
    #191093

    Hicks
    Participant

    I can easily download ( data-dump ) from all 5 systems individually.

    I am looking for a a set of 50 items ( that are all 10×10 ) and comparing each system to one another and within each system.

    I am trying to quantify the differences between the 5 systems.

    0
    #191094

    Hicks
    Participant

    I am thin king about using this Excel worksheet.
    Attribute Gage RR – pass or fail…

    0
    #191096

    French
    Member

    I’m not sure what you expect to get from an Attribute R&R – it just doesn’t seem like the right tool, here.

    It seems the more interesting question is how do the systems get different information. I’m with MBBinWI, Chi-Square will tell you more interesting things like is Oracle more likely to be wrong or a specific type of product, etc. That information is more likely to guide you toward the problems in the processes that caused the information to be incorrect.

    0
    #191097

    Hicks
    Participant

    Here is a picture of the project I am tackling.

    We have many difference systems ( only God knows why ). – I guess that is why they hate hearing.. “I am from the corporate office, and here to help”.

    Corporate Office – K-2, Adage
    Production Plants – Whistle
    3rd Party Freezers ( Red Prairie and High Jump ).

    These databases all hold our specs ( ti-hi, case weight, pallet weight etc.. ).

    My job is to cleanup all 5 systems and verify all 5 have the EXACT same information.
    Considering none of them are connected in any way possible, this will be a very tedious job to do ( over 5000 items ).

    I want to stratify/quantify how screwed up the specs are across the 5 systems, before I start changing specs.

    I was thinking about a pass fail tool..

    if the “master’ spec says 10×10 = 100 cases per pallet.
    I want to calculate some metric of what each system shows before changing them ( to show progress ).

    If one of the 5 systems do not show 10×10 = 100 ( then it is a FAIL ).

    I want to be able to calculate/quantify the variance/( error ) between the 5 systems.

    I hope this makes sense….

    0
    #191125

    Falcon
    Member

    Hmmm…. I’m not sure that discrete is the way you want to go with this data. If so, a Gage R&R would actually be measuring whowhat is applying the PASS or FAIL designation, not the system’s value.

    Will all of the data you are analyzing be continuous ? i.e. will they all have numeric values ? or will there be discrete data such as colors or names, etc ? If you can, you may want to address the data values themselves as long as they are continuous data. Then, your Gage R&R would be testing the values as stored in the systems against the control values.

    Hope this helps !

    0
    #191126

    Hicks
    Participant

    Ok, here is a new spin on what I am trying to accomplish.

    We have 5 databases. None of them are linked. So the data for Items #’s can be in all 5 with differnt specs. or not in any of them.

    Before I fix the problem and clean up the databases, I want to quantify the original problem – with differences between the 5 databases ( against the master info ). Some databases may not even have the item listed, and may have the item listed with incorrect or ‘blank’ data.

    Hope this helps.
    I just want to know how screwed up the situation is before I begin.

    0
    #191134

    Falcon
    Member

    I think I see where you are headed… You aren’t interested in determining if the 5 systems can come up with the same value; you are interested in determining if the 5 systems have the same value and that’s it. Since you aren’t testing the systems capabilities to generate the same data, this isn’t an MSA. You are just analyzing data.

    If this is the case, my suggestion is to simply extract the data from all systems, cross-reference them against your control data, and assign discrete values for matchno-match against the control. Then simply countcalculate the number of no-matches (i.e. defects) for each system to determine your level of “screwed-up-ness”.

    Steve

    0
    #191135

    Hicks
    Participant

    That is correct. Since all 5 systems are not linked, and entered by hand – the possible level of ‘screwed-up-ness’ is great.

    Finally, someone that understand me. ha.

    Thanks.

    0
Viewing 10 posts - 1 through 10 (of 10 total)

You must be logged in to reply to this topic.